| Popis: |
For an accurate performance assessment of a multi-stage compressor, the circumferentially non-uniform flow at the compressor exit needs to be understood and sampled in a way that minimizes uncertainties. To quantify the effect of the measurement rake positions in the exit duct on compressor performance a combined computational and experimental approach is used on a modern 4-stage compressor. The computational analysis is based on unsteady calculations of a 180-degree sector of the test compressor and experimental verification is provided by comparing to area-traverse data downstream of the outlet guide vanes. It is shown that the exit measurement rakes are subject to circumferential flow variations caused primarily by the combined effect of the potential field of the struts housed within the exit duct and the wakes originating from the outlet guide vanes. A circumferential camber pattern, applied to the outlet guide vanes, designed to shield the upstream compressor blade rows against the potential field of the exit struts, is found to reduce the amplitude of the circumferential variation in stagnation pressure and shift its circumferential phase. Recognizing that a smaller numerical model, consisting only of the last rotor, the outlet guide vanes and the exit struts, is sufficient to capture the relevant flow mechanisms, the circumferential variations in stagnation pressure and temperature at the rake position are quantified as a function of the exit capacity. The stagnation pressure and temperature uncertainty within a +/-2 deg circumferential range around the nominal rake position is found to be up to 2.25 times larger than the change of the nominal values over an 87.1–106.0% variation of the exit capacity. Three options to position the rakes to reduce the uncertainty in compressor efficiency are presented — moving the rake downstream as well as leaning and verniering the rakes over the outlet guide vane pitch. Moving the rake from the leading edge to the trailing edge plane of the exit struts reduced the efficiency uncertainty by 2.6%, while leaning and verniering the rakes reduced the efficiency uncertainty by 0.2% and 0.7% respectively. The knowledge gained from the large-scale, detailed CFD predictions can used to support future measurement campaigns. |
